vit-base-aihub_model
This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.0080
- Accuracy: 0.9979
- Precision: 0.9982
- Recall: 0.9976
- F1: 0.9979
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 128
- eval_batch_size: 128
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 512
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1 |
|---|---|---|---|---|---|---|---|
| 0.015 | 1.0 | 101 | 0.0229 | 0.9938 | 0.9928 | 0.9929 | 0.9928 |
| 0.0177 | 2.0 | 203 | 0.0316 | 0.9913 | 0.9928 | 0.9880 | 0.9903 |
| 0.0177 | 3.0 | 304 | 0.0127 | 0.9958 | 0.9954 | 0.9955 | 0.9954 |
| 0.0145 | 4.0 | 406 | 0.0129 | 0.9967 | 0.9970 | 0.9960 | 0.9965 |
| 0.0101 | 4.98 | 505 | 0.0080 | 0.9979 | 0.9982 | 0.9976 | 0.9979 |
Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3